I'm trying to improve my tactics and thinking-ahead on various different websites. Problem is, it's always only a couple of moves ahead. I would really like to find good websites / books / videos dealing with longer combinations.
Ideally they start out 'easy' (couple of steps thinking ahead) and then throughout the exercises get longer and more complicated.
Get better at tactics. Lichess will give you harder problems
It will? I had no idea.
My advice is:
Try out as many sites as possible and find out which site suits you the best.
1. Try Lichess puzzle training
2. Try the 3 different modi (standard, mixed, blitz) on chesstempo.com
3. Try the "puzzle rush" and "puzzle battle on chess (.com)
If your rating improves, all of these sites will give you harder puzzles. This is sometimes counter-productive, because you need more time to solve them.
Be careful with the rating system. Don't take it too seriously. It can be very demotivating when you get punished with -30 points for not solving a puzzle, but knowing that you would only have been rewarded with 1 point, if you had solved it.
Sometimes it is better to log out from your account to play puzzles as an unregistered user. Then you get more easier puzzles and can play faster. This is better for blitz-training or just for fun.
What works also is looking up in YouTube, the St. Louis Chess Club channel for instance, where GMs give lessons. I usually learn stuff about tactics there since I posess no chess books myself. Otherwise, just solve puzzles here. The only thing for now I have problems at finding training for is endgames. But the more consistently you train and learn, the better you become, so simply try that.
The only better option than the suggested by me and the others in this topic most likely would be to get a coach. Like I love saying: Practice makes perfect and sometimes you must also practice with the perfect.
I am not happy with the lichess puzzles, too. Why?
1. In chess practice there is very often a second, even a third solution. Maybe not so fast, maybe not so elegant. But with lichess puzzles there is only one correct solution, for all other possibilities you will be punished. But this is not the truth on the board.
2. The aim of chess is to win. Not to find the golden Michail-Tal-combination! So, a move which leads to a better position should not be punished with a minus of rating.
3. A lot of puzzles like to surprise with a strange or "crazy" combination. Yes, it is interesting, but it's not helpful for the real chess challenge OTB. The result is the seduction to try the most surprising move. This is a bad education.
4. A human player is not an engine. The real opponent does not react like the virtual opponent of the puzzle. How often reacts the virtual opponent - following the engine - with a strange sacrifice which has absolutely nothing to do with the chess reality between human players.
According to this, I suggest: 1. A move which leads to a sudden win: full points. 2. A move which leads at least to a better situation: 75% of points. 3. A move which does not improve the situation but neither leads to a worse situation: 0 points. 4. A move which leads to a worse situation: minus.
I think this would be a fair system. For me, the actual rating system is completely frustrating. To be sincere: I learned to hate it. I lost my joy solving puzzles.
@Immer_um_den_See I agree with you on researching multiples goals of training via the same puzzle (having a more detailed reinforcement system like you suggest would do that)
perhaps this means changing the automatic problem selection filters, they have their own bias, even if automated. Given the increased size of userbase, and probably of daily games database, maybe that is possible, if that was a limiting factor. Because good problems may be defined differently, if multiple facets considered... i am just trying to see why your idea is not already implemented.
I don't see why, even with the current problem selection process*, a less peaked reward system could not be implemented. who cares about backward user rating compatibility, if that is a problem. I rarely compare myself with other from based on puzzle history, because it is more an individual progress measure, i would be glad if i could learn many aspects transposable to general chess games, with the same puzzle, even at the cost of rating reliability.
other objections to a graded reward system?
*: don't know much about it, guessed it was automated, with quantifiable criteria.
The puzzles always have a definite Solution. There is no "second best" or alternative.
"lot of puzzles like to surprise with a strange or "crazy" combination" - what is that assumption based on? The puzzle will either have a move that is by far the best or a move to "hold on". I dont see how this is a bad thing.
I agree with you. Often there are puzzles that look like there's an alternative solution, but when you look a little deeper you see that it doesn't work.